Open AI holds a leading position in AI race despite challenges
he evolution of the industry hinges on whether the demand for general AI models, as exemplified by Open AI and Google, prevails, or if specialized firms concentrating on specific tasks gain prominence, potentially fragmenting the AI landscape.
Open AI, a startup supported by Microsoft, has established itself as a leading player in the race to dominate the future of AI. The company launched its chatbot, Chat GPT, which quickly gained 100 million users, prompting rivals such as Google and Alphabet to release their own chatbots. In the first half of 2023, venture capitalists invested over £40 billion in AI firms. However, public interest in AI has declined, decreasing monthly visits to Chat GPT’s website.
Nevertheless, Open AI maintains a technological advantage. Its latest AI model, GPT-4, surpasses competitors in various benchmarks, including the ability to answer reading and math questions. Open AI is also experiencing revenue growth, earning around £1 billion annually compared to just £28 million before Chat GPT was launched.
Open AI’s funding model, which includes a capped-profit company structure, has attracted investors, raising around £14 billion. Microsoft, through its Azure cloud division, is a major investor and also provides Open AI with the necessary computing power. In return, Microsoft will receive most of Open AI’s profits if distributed. Microsoft also has licensing rights to Open AI’s technology, giving it an advantage in offering AI solutions to its corporate customers.
To maintain its position, Open AI must avoid the fate of previous tech pioneers being overtaken by competitors. As a first mover in the AI industry, Open AI’s decisions will shape the direction of the overall industry. Despite advantages such as scale and high barriers to entry for competitors due to the high fixed costs of GPT-like models, Open AI has yet to witness the network effects that have elevated tech giants like Alphabet, Amazon, and Meta to quasi-monopolistic positions.
Open AI’s focus on experimentation incurs high costs, with many ideas failing before the training stage. Competitors like Anthropic are gaining traction, and customers are seeking alternatives to reduce dependence on Open AI and mitigate the risk of being locked into its products. Concerns regarding data privacy and the black-box nature of Open AI’s models may also deter potential users.
However, Open AI is adapting its strategy to become a business-to-business platform. It develops bespoke products for corporate customers and provides tools for developers to build products using its AI models. The company has a £175 million fund to invest in smaller AI startups building applications on its platform, allowing Open AI to promote its models and capture value from successful applications.
Challenges remain for Open AI to maintain its position in the competitive AI race. Competitors are emerging with powerful models, such as Google’s Gemini, rumoured to be more potent than GPT-4. Even Microsoft, despite its partnership, poses a competitive threat with access to GPT-4’s black box and established relationships with corporate IT departments. The availability of alternatives is diminishing open AI’s pricing power. Addressing transparency and data privacy concerns will be crucial to maintaining its position and continuing to train better models.
The value that comes from a model’s generality will influence the industry’s landscape. If generality is highly valued, big models like those of Open AI and Google may dominate. Alternatively, if specialized firms focusing on specific tasks gain prominence, the industry may become more fragmented.
Source: The Economist